Glossary
The vocabulary of AI-assisted iOS debugging — what each term means and how it relates to giving your AI assistant eyes on a device.
- app-agnostic monitoring
- App-agnostic monitoring observes any application without modifying its code or requiring an embedded SDK. ArgusTest monitors any installed iOS app at the device level — your own builds, client builds, or third-party apps — with no rebuild and no integration step, because it reads device-level signals rather than hooking into the app.
- crash log — .ips crash report
- A crash log (an .ips file on modern iOS) is a structured report Apple generates when an app terminates abnormally, containing the stack trace, thread state, and termination reason. ArgusTest detects crashes automatically and makes the report available to your AI assistant, so it can analyze the stack trace the moment a crash happens.
- device daemon
- A device daemon is a background process that continuously monitors a connected device without manual intervention. ArgusTest's daemon runs on macOS as a LaunchAgent, watching the attached iOS device — capturing screenshots, streaming logs, and detecting crashes — and exposes that data to AI assistants through its MCP server.
- error deduplication
- Error deduplication collapses repeated, identical errors into a single representative entry so the signal isn't drowned by noise. ArgusTest fingerprints errors with SHA-256 signatures and removes duplicates, cutting log noise by 78% before the stream reaches your AI assistant — so the assistant reasons about distinct problems, not thousands of copies.
- iOS device debugging
- iOS device debugging is the process of inspecting a running app on a physical iPhone or iPad — its screen, logs, crashes, and state — to find and fix defects. ArgusTest makes this visibility available to AI assistants, so an assistant can diagnose issues from real device data instead of relying on you to describe them.
- LaunchAgent
- A LaunchAgent is a macOS mechanism (managed by launchd) for running a background process on behalf of a logged-in user. ArgusTest installs as a LaunchAgent so its monitoring daemon starts automatically at login, restarts if it crashes, and keeps running across reboots without you launching it manually.
- MCP — Model Context Protocol
- MCP (Model Context Protocol) is an open standard that lets AI assistants call external tools and read external data through a consistent interface. ArgusTest exposes your iOS device to assistants like Claude Code and Cursor as a set of MCP tools, so the AI can request screenshots, logs, and device state directly.
- MCP server
- An MCP server is a process that exposes tools and resources to AI assistants over the Model Context Protocol. ArgusTest's MCP server runs locally on macOS and surfaces iOS device capabilities — screenshots, log streams, crash reports, device status — as callable tools for Claude Code, Cursor, and other MCP clients.
- offline grace period
- An offline grace period is a window during which licensed software keeps working without contacting its license server. ArgusTest validates your license periodically but allows a 14-day offline grace period, so monitoring keeps running on planes, in secure environments, or through intermittent connectivity.
- wireless debugging
- Wireless debugging means inspecting a device over Wi-Fi instead of a USB cable. ArgusTest captures screenshots and streams logs from a connected iPhone or iPad over the network, so your AI assistant can see the device without it being tethered to your Mac.